Utilization of game along Brazil's transamazon highway

Tbe hunting methods and game yield írom rain forest and second growth along the Transamazon highway are described and the importance of 'Wild meat in the diet of setlers is assessed . The implica tions of game take with respect to the human carrying capacity of interfluvial forest in Amazonia is also explored.


INTRODUCTION
In 1970, Braz i I initiated an ambitious scheme to populate the 'demographic void' of the Amazon basin , which includes the world 's largest tropical rain forest of 5 million sq . km . The pioneer 3,000 km Transamazon highway. completed in 1974. was conceived as the main axis for settlement of the North ( Figure 1) . As of Decembet , 1974, a total of 5,717 families had settled along the highway in 27 agrovilas, cont<:~toing 48 t o 66 houses. 3 agropoli intermediate-sized towns, and 1 ruropolis, an administrative center at the apex of the ruralurban hierarchy (Smith , 1976a) .
Although tropical rain forests are rich in species of both plants and animais, the population of any one species in a given area is usually low . Furthermore, livestock populattons in a pioneer zone require severa! years to build up for sustained yield slaughter . Thus. the contribution of game to the diet of Transamazon settlers is examined in arder to assess its importance in satisfying the protein requirements of settlers. Finally, the implications of game yield with respect to the hum:m carry:ng capacity of the interfluvial forests of the region are investigated .

STUDY AREAS AND FIELD METHODS
lrF 1973-4, the wildlife take of three well separated agrovilas was examined . The region Nigel J. H. Smith (* ) around the first, Coco Chato , at km 42 along the Marabá-Altamira stretch, had been settled previously by some thirty peasant (caboclo) families which had engaged in subsi stence slash and burn agriculture for approximately fifteen years. Thus. mature upland forcst (mata pesada) is interspersed with second growt h at various stages of succession. The second agrovila. Leonardo da Vinci , at km 18 ,L\Itamira-Marabá, is surrounded by rel atively undisturbed liana forest (m ata de cipó, cf . Pires, 1974;Goodland and lrwin, 1975:1 03) . whereas mature upl and forest predominates in the region of the third agrovila, Nova Fronteira. km 80 Altamira-ltaituba. In these study areas, the annual rainfall averages approximately 1 ,700 mm with a pronounced dry season extending f rom June through December. According to the Koppen system . t he climate is classified as Aw (Smith . 19í6b) . Ali ot the study sites are on non-flooded terra firme .
By accompanying hunters and by interviewing ali residents at least every eight weeks within a 5 km radius of the study agro· vilas, the harvest of wildlife was cstim:::lted . A total of 25 hunts were witnessed, 12 ot which registered kills. Colonists were founà to estim:~te the weight of their kills reasonably accurately. Howev er. since settlers are more likely to forget smaller game, the compiled data (Tables 1-3) may be biased in favor of larger species. Only two overnight huntinu trips were conducted from the study agrovil :1s. Most of the settlers hunt within 5 km of t heir agrovilas. The 100 sq. km cropping area surrounding each agrovila was determined bv observing side-roads, built every 5 km aiong the Transamazon in the study zones , and by demarcation paths at the end of the 100 h a lots .

DIURNAL IIUNTING METHODS
Wíthín each communíty, two to four hunters (caçadores) normally make at least one tríp a week into the forest in pursuit of large game and account for most of the game take by weíght. In addition, during the course of agricultura! work settlers kill wild animais, especially smaller specíes such as agoui:l~ (Dasyprocta sp.) and níne-banded armadillos (Dasypus novemcinctus) . Most hunters work alone, but two or more were present 3t 24% of the recorded ki lls. Men may form a small group íf there are sígns of tapir (Tapirus terrestris) or whítelípped peccaries (Tayassu pecari) in the area. Some hunters consíder dogs 3 nuísance and do not use them; others always take one to four of them into the forest to help in locating game. Dogs were present at 34% of the kills. Leaves of Piper lanceolatum (Piperaceae) are sometimes rubbed into tne noses of dogs, apparently to sharpen their sense of smell and thus ímprove their trackíng ability. Dogs usually fan out in front of the hunter and thus cover a comparatively large area of forest. The caçador contrais his dogs with loud calls, and the dogs indicate th~ presence and kind of game with yelps and barks.
lf the dogs locate a band of white-lipped peccaries, the latter often flee, in which case the hunter may not get close enough for a shot. But sometimes the band, which may contain 100 members, turns to fight the dogs, and thereby allows the hunter to catch up and kill as many as eight peccaries, usually with a 16 o r 20 gauge shotgun. In the fight, however, he may lose severa! dogs to the slashing peccary tusks. Without dogs, the hunter may encounter undisturbed white-lipped peccaries in a mud wallow, or feeding on fallen palm fruits of açai (Euterpe oleracea) or paxiúba (/riartea sp.) by a stream. The clacking of peccary teeth may _ warn the hunter at a distance of 200 meters, so that he can approach his quarry from downwind. Tracking a band of porcão, as T. pecarí is known in the study areas, is relPtively easy since it produces a broacl wake of shuffled Jeaves and tracks.

456-
viduals, often flee from dogs by hiding inside a hollow log or an armadilio burrow as small as 25 cms. in diameter. I f the peccaries are inside a rotting trunk, the hunter uses a machete to gain access to his prey. I f the quarry has entered a hole in the ground the hunter drives st3kes across the entrance and pokes a long, thin pole into the tunnel to irntate caitetu, as T. tajacu is known in the Brazilian Amazon. When a peccary comes to the surface, the hunter easily kills it at the barrier with a shotgun blast. H e then removes the carcass, replaces the stakes, and jams the pole in again in an attempt to locate other peccaries. Up to four have been killed trom a single burrow in this manner . rap1rs, weighing up to 200 kg, art: tne largest mammals of the terra firme forests oi Amazoni a. Hunters find them by toliowing their distinctive footprints and occasiona 1, unmistakabiy-large teces piles. Somenmes, the shnll wnistle of a tapir aierts the hunter , who then approaches with extreme caut10n from downwind. I f dogs pursue a tap1r, it usua11y turns to fight in a stream where 1 t h as an advantage, frequently inflicting severe DltE::ti on the dogs. While the tapir is thus occupied, the hunter is able to dispatch his prey witn several shotgun blasts.
Some hunters train dogs to locate brocket deer (Mazama americana) and agoutis and to chase the game to them. Only rarely are the dogs alone able to bring down a full-gl'own brocket deer of about 50 kg. After shooting a deer, the stomach contents of the kil! are rubbt:·d into the hair of the dogs, apparently to make them enemies of deer and eager trackers.
At dawn and dusk, colonists hide in forc ::;, or second growth at the edge of maize fields for agoutis to come and feed. Settlers ais o wait for agoutis near babaçu palm (Orbygnia marttana) since the pulpy mesocarp of the fistsized fruit is a favored food of the rodent (Smith, 1974). Agouttis favor areas of s1as11 and burn agriculture where the fire-reststant palm proliferates. Fire liberates the oily nuts in the endocarp and these germinate readtly in the nutrient-rich ash. This is an exceptional case, however, since most of the major game [mimais along the highway, such as tapir and white-lipped peccary, are eliminated or considerably reduced when their forest habitat is àestroyed.

NOCTURNAL HUNTING METHODS
Night hunting is largely confined to the dry season in order to avoid being drenched, and when the rustle of leaves or the sn3p of a twig betrays game. In the espera or waiting method, a man slings a hammock or constructs a narrow platform some two meters above ground within shotgun range of a tree with fallen fruit or flowers and signs of feeding by game. The hunter arrives before sundown and may wait untd dawn, shotgun and flashlignt ready, ignoring insect bites. Forest trees with falling fruit much sought by brocket deer Hagh~oo oys ASPHALT UNPAVED Plonned Borders Rtvers include tatajuba (Bagassa guianensis, Moraceae), gameleira (Ficus spp., Leguminosae). jatobá (Hymenaea spp., Leguminosae), tamboril (Enterolobium maximum, Leguminosae) and frutão (Pouteria pariry, Sapotaceae). Hunters also wait at night for deer in mature forest by trees with falling flowers such as sapucaia ( Lecythis usitata, Lecythidaceae). castanheira (Bertholletia excelsa, Lecythidaceae) and jarana (Ho/opyxidium jarana, Lecythidaceae) .
In one night a hunter may kill one or two pacas (Agouti paca) or a tapir as they feed on fallen gameleira fruit . Pacas are also shot at night when they come out of burrows or hollow logs to feed on dropped flowers of sapucaia, mata mata (Eschweilera spp., LegumimJs3e) or p1quiá (Caryocar villosum, Caryocaraceae) Utilization o f game . .. in forest, fallen fruits of babaçu palm in second growth, or maize in cultivated fie lds. Ninebanded armadillos are taken at nigth while feeding on fruit under najá palms (Maximilliana regia).
The fachear or flashlight method is used to capture armadillos or pacas. lf the former is the quarry, a hunter, usually a small boy, searches the ground in forest or old second growth with dogs, flashlight, and machete. When an armadillo runs, the dogs often capture and kill the animal before it can escape down a burrow. lf one enters a hole, it is dug out, a process that can take an hour or more. To capture pacas by the fachear method, a hunter without dogs follows a stream and periodicuily scans the banks with his flash11ght. lf the caçador spots the red-hued reflection of large paca eyes, he quietly aims his snotgun at the mesmerized rodent.

TRAPs
The hunting technology employed by Transamazon settlers is not as elaborate as that ot some aborígines (Ryden, 1950;Ruddle;, 1970) . Colonists set gun traps for paca and nine-banded armadillos only on individual lots f ar from agrovi las . No pitfalls were employed in the study areas. Some settlers built box traps with a collapsable roof or sliding door to capture ocelots (Felis pardalis), since a good pelt from that spotted cat brought the hunter about U. S. $40 in 1974S. $40 in (Smith, 1976c. Sometimes, hunters with dogs tree a wild cat in daylight and shoot it. In spite of Brazilian law 5.197/67, that prohibits commerce in wild animais or their products, skins from two jaguars (Fe/is onca) and three ocelots were sold by hunters in the vicínity of agrovila Nova Fronteira and Leonardo da Vinci in 1974. Although the coats of black jaguars and pumas (Felis concolor) have little market value, their meat is appreciated.

GAME YIELDS
A total of 3,214 kg of game was taken within the 1 O~ sq. km hunting area of agrovi la Nova Fronteira during the course of twelve months in 1973/ 4 (Trble 1). Allowing for an average weight loss of 40% due to discarded 458-bones, visc. era and skin (cf. White, 1953), the actual amount of game meat consumed was about 1,928 kg. A l lowing for variation among species, the protein content of the total game take is probably close to 20%. Thus, the hunfing yield could supply each of the 204 agrovila members with 5 g protein a day.
The average protein requirement of an agrovila resident ís calculated as O. 7 g animal protein/kg body weight/day, based on reference weights emp loyed by W. H. O. ( 1973: 100), and population surveys of the study communities conducted by the author which reveal that 43%, 49% and 51% of the inhabitants of agrovilas Leonardo da Vinci, Nova Fronteira, and Coco Chato respectively are under 15 years old. Thus, the game take for agrovila Nova Fronteira could supply only 17% of the protein needs of the community.
Settlers in the vicinity of Leonardo da Vinci, the other study agrovi la surrounded by relatively undisturbed forest, killed an estimated 3,389 kg of game during 1974 (Table 2). Allowing for 40% wastage, approximately 2,033 kg of game was consumed by the 179 agrovila residents, enough meat to supply about 20% of the protein needs of the community. The much smaller game yield from the vicinity of agrovila Coco Chato (Table  3) is attributed to 15 years of hunting and habitat alteration.
Tapirs and white-llpped peccaries, two of the most important species in forested areas in terms of game take by weight, are no longer found within 5 km of the commun ity. The 761 kg of undressed game could supply each of the 351 members ot agrovila Coco Chato with only 2% of their protein requirements.
Although game taken in predominately forested areas is sufficient, in theory, to provide approximately 18% of the protein needs of the agrovi la populations, whether each individual receives a significant amount of game depends on a number of ecological and cwtural variables. Both the temporal and spatial distribution of kil ls, for example, depend on the season and cultural background of the settler.
The yield of game from forested areas in the vicinity of agrovi las Nova Fronteira and Leonardo da Vinci fluctuates markedly during the year (Figure 2) . The yield is greatest during the latter part of the rainy season, primarily because of the increased kil l o, white-lipped peccaries. Seasonal differences in the number of hunting trips cannot exp lain the greatly increased game yield in the wet season, since settlers have ample opportunities to hunt during the relatively slack months towards the end of the dry season after fie lds have been cleared. One factor is that, according to hunters, porcão is more active in the rainy season because more forest trees bear fruit and because the season:3lly-dry streams are flowing. Game take is lowest during the early part of the rainy season (December to January) partly because colonists are then occupied in planting crops. The temporal variation in game yie ld is less marked in the more disturbed environment surrounding agrovila Coco Chato. There the two most important game animais, agoutis and nine-banded armadil los, are captured throughout the year . The increased game take during the dry season ( Figure 2) is attributed m3inly to nocturnal kills of deer by the espera method.
Even during the months of low yield, game contributes significantly to the diet, not only in terms of supplying protein, but also for the psychological benefit. Settlers consider lunch and dinner unsatisfying without some form of meat or fish . lt is common practice to salt and sun-dry some game meat, especially large anima is such as tapir and brocket deer. for consumption during the months of low hunting yieid.
The distribution of game within Transamazon communities is also uneven. Unlike many aborígines, agrovila residents do not share thei r kills equitably. Along the highway, the distribution of game is confined to hunters and their immediate families, or to those who can afford to purchase game meat at U. S. $0.70/ kg . Alternatively , depending on the needs of the hunter, other settlers may barter goods, such as rice, man ioc flour, or tobacco, in exchange for meat.
Another obstacle to the widespread distribution of game within agrovilas is the system of food avo id 3nces wh ich reduces the consumption of certai n species. Southerners, for example, tend te avoid unfamiliar game . Some gaúchos and Paranaenses consider the meat of most forest animais unclean . More familiar animais. such as .agoutis, pacas and rabbits , which occur in southern Brazil , are, however, more acceptable. On the other hand , Nordestinos and Amazonian peasants shun rabbit meat as too slippery (lisa). Settlers who feel ill generally avoid " strong " or reimosa game (Table 4), thinking i t will provoke latent disease or exacerbate symptoms of a chronic illness. Tapir meat, for example , is deemed especially offensive to the liver, while ingesting succulent paca flesh can apparently transform a cold into a case of pneumonia . The fatty meat of armadil los . tortoises (Geochelone sp.) and col lared peccary is considered prejudici al to those susceptib le to skin disorders. The origin of the food avoidances of Transamazon settlers is unclear, but ma ny have undoubtedly been absorbed , in modified form, from aboriginal cultures. Among the Witoto of the Braz i lian northwest, for example, tapir meat is considered to be very strong, especial ly for women , and is thus eaten sparingly (Whiffen, 1915:126) . Females of Kayapó groups, which inhabit the interfluvíal forest between the Tocantins and Xingu , shun torto ise meat since it is thought to inflame the skin (Dreyfus, 1963).

FOR SETTLEMENT
The interfluvial forests of Amazonia, occupying approximate ly 95 % of the basin, are generally considered to be inhospitable for     hunters, gatherers, and subsistence farmers because of the limited fish resource, sparse game populations, and generally poor soils (Castro and Reis, 1952;Meggers, 1954;Ferdon, 1959;Denevan, 1968;Lathrap, 1968;Denevan, 1970;Meggers, 1971;Gross, 1975). For example, Galvão (1963) claims that aboriginal groups cannot attain a size larger than 200 individuais emp loying manioc (Manihot escu/enta) as a basic staple on the terra firme, although Carneiro (1960Carneiro ( , 1961 has convincingly argued that a tribe of 2,000 can be supported on a completely sedentary bas1s in upland forests of Amazonia by cultivating the root crop in a swidden system within 6. 4 km of a village. Although the potential of manioc as an abundant and reliable source of energy is generaly well-recognized, severa! writers have suggested that a shortsge of protein in the terra firme forest is a major factor limit ing the size and permanence of aboriginal groups (Denevan, 1!166 Carneiro, 1970a;Denevan, 1971;Gross, 1975). In spite of the fact that game provides only a fraction of the protein needs of Transamazon colonists in areas previously occupied for as little as fifteen years by subsistence farmers, and that Amazonian tribes do not rear domestic animais for food ( 1 ), the low density of mammalian populations in Neotropical rain forests (cf. Eisenberg and Thorington , 1973;Fittkau and Klinge, 1973) does not preclude the possibility of large sedentary settlements in non-riverine environments of Amazonia .
The generally small size and semi-nomadic habits of groups currently occupying the interfluvial forests of Amazonia may be attributed to a number of factors not related to protein shortage o r soi I exhaustion but to introduced diseases (Neel, 1970;Arnaud andAlves, 1974, Arnaud, 1975) ( fiss ion as a result of internai disputes (Lima, 1950;Goldman, 1966;Chagnon, 1976) , warfare, and superstition. Apparently, the natural resources of terra firme forests are capable of supporting large, ( 1 ) -However, Amazonian lndians keep monkeyis and peccaries as pets and raise parrots, macaws and harpy eagles (Harpia harpyja) for their feathers which are employd in ceremonial wear (Gilmore, 1950;Carvalho, 1951) . The fact that no Amazonian tribes domesticated animais for meat production may reflect the abundance o f wild sources of animal protein.
sedentary native settlements. In 1824, for example, even after a century of intermittent contact with Luso-Brazilians and at least one smallpox epidemic, three villages of the Apinayé, who still inhabit in much reduced numbers the interfluvial region between the Tocantins and Araguaia, contained a 1,000 individuais or more, with the largest sheltering 1,600 inhabitants (Nimuendaju, 1939).
The fact that 23 archaeological sites are situated in the study areas, 4 of them associated with anthropogenic black earth (terra preta do indio), suggest that sedentary aboriginal groups were able to occupy terra firme sites along the Transamazon without exhausting soil or protein resources. The four terra preta sites range from O. 5 to 5 ha in area, and assuming that the groups lived in communa! houses, from 100 to at least 500 inhabitants may have occupied the sites. No stratigraphic studies of any Transamazon archaeological sites has been made so it is not known of they were occupied continuously. However, a preliminary analysis of potsherds collected by the author (DeBoer, et ai. 1976) suggests that at least 2 of the black earth sites were occupied by one group (single ware) . Further, since 87 cms. depth of terra preta messured at one of the sites probably represents close to a century of kitchen middens and ash accumulation (cf. Evans, 1964). it seems plausible that relatively large groups occupied upland sites along the highway transect without destroying soil or protein resources.
A number of cultural traits may partly explain how the native groups were able to crop wild sources of animal protein without exhausting the resource. The concept of game may differ among cultures occupying the same region. Whereas mammais account for 96% of the game take around the highway agrovilas (Tables 1-3), abotriginal groups of Amazonia generally exploit a much wider spectrum of the animal biomass. lnsects, an excellent source of high quality protein (Piatt, 1962;FAO, 1970). account for most of the animal biomass in the terra firme forests of central Amazonia (Fittkau and Klinge, 1973), and are widely consumed by Amerind groups (Carvalho, 1951;Oberg, 1953;Bruzzi, 1962:220; Utilization of garoe . .. Chagnon, 1968:30;Baldus , 1970:165;Ruddle , 1973;Reichei-Dolmatoff, 1974:62;Taylor, 1974: 23) . However, Transamazon colonists were never observed eating arthropods.
lnstead of cropping more abundant mammals, such as rodents and marsupiais, settlers in forested areas along the Transamazon concentrate on relatively large taxa such as peccary, tapir and brocket deer, the latter three species compose 89 % of the game take by weight in the vicinity of agrovilas Nova Fronteira and Leonardo da Vinci (Tables 1 ,2) . On ly when the larger species become sc3rce in heavily hunted and modified habitats, are rodents significantly represented in the annual kill. Thus pacas and agoutis account for 39% of the game take in the 100 sq. km cropping area around agrovila Coco Chato , where the habitat h as been degraded, but rodents account for on ly 3 % of the annual game take by weight in predominately forested areas. However, rats and mice, notorious agricultura! pests along the highway, and a source of food for some aborigina l groups (Wilbert. 1974), are not eaten by Transamazon colonists.
Wild birds, a minor source of anima l protein to co lonists in the study areas, account fcr on ly O. 6% of the game yield by weight in forested regions, and 2. 6% in the vicinity of Coco Chato (Tables 1-3), whereas in some Amerind groups, birds are an important food (Rudd le, 1970). By hunting and gathering a diverse range of animal species, and by taking advantage of abundant taxa such as rodents and arthropods, pressure on game is more evenly distributed and a more reliable protein supp ly is assured.
Aborígines crop larger areas than agrovila residents and thus conserve local game populations. For example, Gê groups, such as the northern Kay::~pó of the middie and upper Xingu (Dreyfus, 1963). and the Xikrin of the ltacaiLInas river in southeastern Pará (Caron, 1971: 278) conduct extended hunting trips lasting from severa I days to a month or more. Non-Gê groups such as the Maracá of eastern Colombia (Ruddle, 1970) may also hunt for severa i days at a time. Some of the game taken on such trips is preserved by smoking and sun-drying for later consumption in the village.
Another cultural mechanism which may help to explain how relatively large groups were able to remain sedentary on the terra firme is the strict system of taboos which check the consumption of certain game animais in many native cultures. Some tribes prohibit the eating of some animais during puberty (Levi-Strauss, 1948;Metraux, 1948), menstruation (Nimuendaju, 1948a, gestation (Nimuendaju, 1948a) and immediately following birth (Metraux, 1948: Nimuendaju, 1948a . Such checks may alleviate hunting intensity. Respect for supernatural animal spirits is wide-spread among native cultures of South America (Zerries, 1954;Murphy, 1958:14;Reichei-Dolmatoff, 1974:85;Wilbert, 1974) ancJ helps to prevent the over-exploitation of game, but no such checks operate among hunters in the study areas. Whereas a few settlers along the highway assert that a supernatural humanoid, known locally as pai da caça or caipora, lurks in the forest and may punish anyone who kills too many of a single species, none of the caçadores in the study areas believe that caipora has any power over them. Although peasants tear spirit-protectors of game in other areas of Amazonia (Wagley, 1967:235;Moran, 1974), the traditions of caboclos in the study areas have been diluted by contact with colonists from other regions, particularly from the more developed South.
Another trait that reduces the threat of protein malnutrition is the tradition of sharing kills, especially large ones, among tribal members (Nimuendaju, 1952:32;Leacock. 1964;Frikel, 1968;Carneiro, 1970b;Taylor, 1974:32). However, in highly acculturated tribes such traditions may break down thereby restricting the distribution of meat (Maybury. Lewis, 1956), as is the case with caboclo culture along the Transamazon.

CONCLUSION
A combination of cultural attributes, from consumption of unorthodox game and vegetable sources ~f protein , to the cropping of large areas. food taboos, respect for supernatural protectors of game, and sharing kills among community members, may have enabled groups 464-that formerly occupied the archaeologic31 sites along the Transamazon to remain sedentary without exhausting game populations. A comparison between the cultural traits of aboriginal groups and those of the present settlers along the highway, suggests that the low game intake of the latter is in large measure due to cultural, rather than ecological , factors.

ACKNOWLEDGEMENTS
Field work on the Transamazon was conducted from July to October, 1971, October to November, 1972, and from August, 1973to November, 1974